Example #1
0
def generate_remote_files_from_manifest(remote_file_manifest,
                                        algs,
                                        strict=False):
    logger.info("Generating remote file references from %s" %
                remote_file_manifest)
    remote_files = dict()
    with open(remote_file_manifest, "r") as rfm_in:
        line = rfm_in.readline().lstrip()
        rfm_in.seek(0)
        is_json_stream = False
        if line.startswith('{'):
            fetch = rfm_in
            is_json_stream = True
        else:
            fetch = json.load(rfm_in, object_pairs_hook=OrderedDict)
        try:
            for entry in fetch:
                if is_json_stream:
                    entry = json.loads(entry, object_pairs_hook=OrderedDict)

                filename = ''.join(['data', '/', entry['filename']])
                url = entry['url'][0] if isinstance(entry['url'],
                                                    list) else entry['url']
                hash_provided = (bdbagit.CHECKSUM_ALGOS -
                                 set(entry.keys())) != bdbagit.CHECKSUM_ALGOS
                if not hash_provided:
                    raise ValueError(
                        "A remote file manifest entry did not provide a required hash value: %s"
                        % json.dumps(entry))
                add = True
                for alg in bdbagit.CHECKSUM_ALGOS:
                    if alg in entry:
                        if strict and alg not in algs:
                            add = False
                        if add:
                            bdbagit.make_remote_file_entry(
                                remote_files, filename, url, entry['length'],
                                alg, entry[alg])
        finally:
            rfm_in.close()

    return remote_files
Example #2
0
def generate_remote_files_from_manifest(remote_file_manifest,
                                        algs,
                                        strict=False):
    logger.info("Generating remote file references from %s" %
                remote_file_manifest)
    remote_files = dict()
    with open(remote_file_manifest, "r") as rfm_in:
        line = rfm_in.readline().lstrip()
        rfm_in.seek(0)
        is_json_stream = False
        if line.startswith('{'):
            fetch = rfm_in
            is_json_stream = True
        else:
            fetch = json.load(rfm_in, object_pairs_hook=OrderedDict)

        for entry in fetch:
            if is_json_stream:
                entry = json.loads(entry, object_pairs_hook=OrderedDict)

            entry['filename'] = ''.join(['data', '/', entry['filename']])

            add = True
            for alg in bdbagit.CHECKSUM_ALGOS:
                if alg in entry:
                    if strict and alg not in algs:
                        add = False
                    if add:
                        bdbagit.make_remote_file_entry(remote_files,
                                                       entry['filename'],
                                                       entry['url'],
                                                       entry['length'], alg,
                                                       entry[alg])

        rfm_in.close()

    return remote_files